Fix bugs

run.py

@@ -66,7 +66,7 @@ def config() -> argparse.Namespace:
             "screenshot_a11y_tree",
             "som"
         ],
-        default="a11y_tree",
+        default="som",
         help="Observation type",
     )
     parser.add_argument("--screen_width", type=int, default=1920)
@@ -146,6 +146,7 @@ def test(
             step_idx = 0
             env.controller.start_recording()
 
+            # todo: update max running time for each example, @xiaochuan
             while not done and step_idx < max_steps:
                 actions = agent.predict(
                     instruction,
@@ -158,7 +159,7 @@ def test(
                     action_timestamp = datetime.datetime.now().strftime("%Y%m%d@%H%M%S")
                     logger.info("Step %d: %s", step_idx + 1, action)
 
-                    observation, reward, done, info = env.step(action, args.sleep_after_execution)
+                    obs, reward, done, info = env.step(action, args.sleep_after_execution)
 
                     logger.info("Reward: %.2f", reward)
                     logger.info("Done: %s", done)
@@ -167,7 +168,7 @@ def test(
                     # Save screenshot and trajectory information
                     with open(os.path.join(example_result_dir, f"step_{step_idx + 1}_{action_timestamp}.png"),
                               "wb") as _f:
-                        with open(observation['screenshot'], "rb") as __f:
+                        with open(obs['screenshot'], "rb") as __f:
                             screenshot = __f.read()
                         _f.write(screenshot)
 
@@ -186,22 +187,24 @@ def test(
                     if done:
                         logger.info("The episode is done.")
                         break
-
-            result = env.evaluate()
+            try:
+                result = env.evaluate()
+            except Exception as e:
+                logger.error(f"Error in evaluating the example {example_id}: {e}")
+                result = 0.0
             logger.info("Result: %.2f", result)
-            scores.append(result)
             env.controller.end_recording(os.path.join(example_result_dir, "recording.mp4"))
+            scores.append(result)
+            with open(os.path.join(example_result_dir, "result.txt"), "w", encoding="utf-8") as f:
+                f.write(f"{result}\n")
 
     env.close()
     logger.info(f"Average score: {sum(scores) / len(scores)}")
 
 
-def get_unfinished(test_file_list, result_dir):
-    finished = []
-    for domain in os.listdir(result_dir):
-        for example_id in os.listdir(os.path.join(result_dir, domain)):
-            finished.append(f"{domain}/{example_id}")
-    return [x for x in test_file_list if x not in finished]
+def get_unfinished(test, result_dir):
+    # todo @xiaochuan
+    pass
 
 
 if __name__ == '__main__':